基于高斯混合模型的心音分类研究
Gaussian Mixture Model-Based Diagnostic Research for Combined Heart Diseases Shuping Sun, Tingting Huang, Yarui Pan, Biqiang Zhang, Baojin Liu, Jie Wu
摘要: 为表示复合心脏病的复杂诊断信息，本研究提出一种基于高斯混合模型结合概率诊断心脏病的方法。本文结构如下：首先，利用电子听诊器采集心音信号；其次，采用小波分解对心音信号进行预处理，以保留心音信号有效成分，然后采用功率谱分析结合阈值线方法提取心音信号频率域特征；最后，基于精度建立最优高斯混合模型数量并结合后验概率得出诊断结果。为验证本研究有效性，以2560秒心音数据作为研究对象，实验结果表明，该方法可以概率模式描述不同类别所属程度得出最终诊断结果。
Abstract: To express the diagnosis information for combined heart diseases by diagnosing heart sound, this study proposes a Gaussian mixture model (GMM)-based classification method combined with the probability diagnostic results to diagnose heart diseases. This paper is organized as follows. Firstly, heart sound is collected using an electronic stethoscope. And then, wavelet decomposition is employed to preprocess the heart sound signal and retain the effective components of the heart sound signal, and the power spectrum density method combined with threshold line method is proposed to extract the features for the heart sound signal in the frequency-domain. Finally, the optimal GMM is determined based on the highest accuracy, and the posterior probabilities are proposed to express the diagnostic results. The performance evaluation is verified by the 2560 seconds sounds, and the research results show that the sounds with complex heart diseases can be directly recognized using the different probabilities to describe which kind of heart disease it belongs to.
文章引用：孙树平, 黄婷婷, 潘亚瑞, 张弼强, 刘保进, 吴杰. 基于高斯混合模型的心音分类研究[J]. 计算机科学与应用, 2020, 10(3): 483-492. https://doi.org/10.12677/CSA.2020.103051

参考文献

[1] Hu, S.S., Gao, R.L., Liu, L.S., et al. (2019) Summary of the 2018 Report on Cardiovascular Diseases in China. Chinese Circulation Journal, 34, 209-220.
[2] Sun, S., Wang, H., Cheng, C., Chang, Z. and Huang, D. (2017) PCA-Based Heart Sound Feature Generation for a Ventricular Septal Defect Discrimination. ICCWAMTIP 2017, Vol. 2018, 128-133. [Google Scholar] [CrossRef] 
[3] Chen, T.-E., et al. (2017) S1 and S2 Recognition Using Deep Neural Networks. IEEE Transactions on Biomedical Engineering, 64, 372-380. [Google Scholar] [CrossRef] 
[4] Potes, C., Parvaneh, S., Rahman, A. and Conroy, B. (2017) Ensemble of Feature: Based and Deep Learning: Based Classifiers for Detection of Abnormal Heart Sounds. Computers in Cardiology Conference, Vol. 43, 621-624. [Google Scholar] [CrossRef] 
[5] Kao, W.C., Wei, C.C., Liu, J.J. and Hsiao, P.Y. (2009) Auto-matic Heart Sound Analysis with Short-Time Fourier Transform and Support Vector Machines. Midwest Symposium on Circuits and Systems, Cancun, 2-5 August 2009, 188-191. [Google Scholar] [CrossRef] 
[6] Chen, T.E., et al. (2017) S1 and S2 Heart Sound Recognition Using Deep Neural Networks. IEEE Transactions on Biomedical Engineering, 64, 372-380. [Google Scholar] [CrossRef] 
[7] Stasis, A.C., Loukis, E.N., Pavlopoulos, S.A. and Koutsouris, D. (2003) Using Decision Tree Algorithms as a Basis for a Heart Sound Diagnosis Decision Support System. Infor-mation Technology Applications in Biomedicine, Vol. 2003, 354-357. [Google Scholar] [CrossRef] 
[8] Beritelli, F. and Spadaccini, A. (2010) An Improved Biometric Identification System Based on Heart Sounds and Gaussian Mixture Models. IEEE Workshop on Biometric Measure-ments and Systems for Security and Medical Applications, Taranto, 9 September 2010, 31-35. [Google Scholar] [CrossRef] 
[9] Reynolds, D. (2015) Gaussian Mixture Models BT. In: Li, S.Z. and Jain, A.K., Eds., Encyclopedia of Biometrics, Springer US, Boston, 827-832. [Google Scholar] [CrossRef] 
[10] Nagraniy, A., Chungy, J.S. and Zisserman, A. (2017) VoxCeleb: A Large-Scale Speaker Identification Dataset. Annual Conference of the International Speech Communica-tion Association, Vol. 2017, 2616-2620. [Google Scholar] [CrossRef] 
[11] Sun, S. (2015) An Innovative Intelligent System Based on Automatic Diagnostic Feature Extraction for Diagnosing Heart Diseases. Knowledge-Based Systems, 75, 224-238. [Google Scholar] [CrossRef] 
[12] Varghees, V.N. and Ramachandran, K.I. (2014) A Novel Heart Sound Activity Detection Framework for Automated Heart Sound Analysis. Biomedical Signal Processing and Control, 13, 174-188. [Google Scholar] [CrossRef] 
[13] Jain, P.K., Tiwari, A.K. and Chourasia, V.S. (2016) Performance Analysis of Seismocardiography for Heart Sound Signal Recording in Noisy Scenarios. Journal of Medical Engineering & Technology, 40, 106-118. [Google Scholar] [CrossRef] [PubMed]
[14] 邱鹏, 张建德, 霍瑛. 基于小波分解与神经网络结合的图像压缩算法[J]. 电脑知识与技术, 2018, 14(26): 169-170+174.
[15] 袁向琳. 心音信号的合成模型及特征分析[J]. 心音分析技术, 2018, 11(27): 169-210 + 103.
[16] 王杰, 李蓉, 黄惠东. 基于小波系数的粘连信号穿层特征提取方法[J]. 探测与控制学报, 2016, 38(1): 13-17, 23.
[17] 武肖梦. 基于SSA和EMD的心音频谱特性分析及识别研究[J]. 心音分析技术, 2016, 13(27): 170-180 + 133.
[18] 郑伊能. 基于心音的慢性心力衰竭分期诊断方法研究[D]: [博士学位论文]. 重庆: 重庆大学, 2017.
[19] Zhu, Z., Hu, J. and Wang, Y. (2016) A 0.45 V, Nano-Watt 0.033% Line Sensitivity MOSFET-Only Sub-Threshold Voltage Reference with No Amplifiers. IEEE Transactions on Circuits and Systems I: Regular Papers, 63, 1370-1380. [Google Scholar] [CrossRef] 
[20] Klimont, A., et al. (2016) Low-Threshold, Single-Mode Defect Line Terahertz Quantum Cascade Laser. CLEO: Science and Innovations. Optical Society of America, 2016.
[21] Matrix, C. and Estimation, M.L. (2005) Gmm: 高斯混合模型. 1-8.